SMTPD
收藏资源简介:
SMTPD是由杭州电子科技大学、天津大学和北京大学的研究人员构建的一个新的社交媒体时序流行度预测基准数据集。该数据集通过观察主流社交媒体平台YouTube的多语言多模态内容,收集了超过282,000个样本,主要关注样本发布后30天内的流行度演变。数据集包含了视觉内容、文本内容、元数据和用户资料等多模态信息,并针对每种模态采用了不同的特征提取方法,以进行时序流行度预测。
SMTPD is a novel benchmark dataset for social media temporal popularity prediction, developed by researchers from Hangzhou Dianzi University, Tianjin University, and Peking University. This dataset collects over 282,000 samples by capturing multilingual and multimodal content from the mainstream social media platform YouTube, with a primary focus on the popularity evolution within 30 days following the release of each sample. The dataset encompasses multimodal information including visual content, text content, metadata, and user profiles, and employs distinct feature extraction methods for each modality to facilitate temporal popularity prediction tasks.

- 1SMTPD: A New Benchmark for Temporal Prediction of Social Media Popularity杭州电子科技大学, 天津大学, 北京大学 · 2025年



